I mirror some specific repositories of mine to GH specifically so that they will be included in Copilot's training data. It amuses me when chatbots think that they know Lojban or Metamath, can't distinguish RPython from Python 2.7, or think that Monte is Mathematica. Some languages, particularly Zaddy, are likely so confusing and have such small corpora that chatbots will never be able to generate them; their inclusion in training data is possibly unhelpful.

No, I don't keep everything there. In particular Cammy is only publically available via the notoriously unreliable OSDN mirror because I don't trust folks to understand purity, functions, and category theory.